International Journal of Recent Technology and Engineering (IJRTE) ISSN: 2277-3878, Volume-8 Issue-4, November 2019 4887 Published By: Blue Eyes Intelligence Engineering & Sciences Publication Retrieval Number: D8362118419/2019©BEIESP DOI:10.35940/ijrte.D8362.118419 Machine Learning Based Classification Models for Financial Crisis Prediction S. Anand Christy, R.Arunkumar Abstract: Financial Crisis Prediction (FCP) being the most complicated and expected problem to be solved from the context of corporate organization, small scale to large scale industries, investors, bank organizations and government agencies, it is important to design a framework to determine a methodology that will reveal a solution for early prediction of the Financial Crisis Prediction (FCP). Earlier methods are reviewed through the various works in statistical techniques applied to solve the problem. However, it is not sufficient to predict the results with much more intelligence and automated manner. The major objective of this paper is to enhance the early prediction of Financial Crisis in any organization based on machine learning models like Multilayer Perceptron, Radial basis Function (RBF) Network, Logistic regression and Deep Learning methods and conduct a comparative analysis of them to determine the best methods for Financial Crisis Prediction (FDP). The testing is conducted with globalized benchmark datasets namely German dataset, Weislaw dataset and Polish Dataset. The testing is performed in both WEKA and Rapid Miner Framework design and obtained with accuracies and other performance measures like False Positive Rate (FPR), False Negative Rate (FNR), Precision, Recall, F-score and Kappa that would determine the best result from specific algorithm that will intelligently identify the financial crisis before it actually occurs in an organization. The results achieved the algorithms DL, MLP, LR and RBF Network with accuracies 96%, 72.10%, 75.20% and 74% on German Dataset, 91.25%, 85.83%, 83.75% and 73.75% on Weislaw dataset, 99.70%, 96.30%, 96.21% and 96.14 on Polish dataset respectively. It is evident from all the predictive results and the analytics in Rapid Miner that Deep Learning (DL) is the best classifier and performer among other machine learners and classifiers. This method will enhance the future predictions and would provide efficient solutions for financial crisis predictions. Keywords: Financial Crisis Prediction; Machine learning; Artificial intelligence; Deep learning I. INTRODUCTION Financial companies, corporate, borrowing firms as well as government agencies urge to design models to effectively investigate the possibility of counterparty default. Though default actions act in a stochastic manner, financial data can be employed to design financial crisis prediction (FCP) models. For instance, [1], applied the multivariate statistic methodologies basically, discriminant analysis for classifying solvent and insolvent companies by exploiting financial data. The financial crisis happens not only because of bankruptcy and also due to the degrading of debt ratings of credit- related properties. Revised Manuscript Received on November 15, 2019 * Correspondence Author S. Anand Christy*, Department of Computer and Information Science, Annamalai University, Chidambaram, Cuddalore, TamilNadu, India. Email: scholarchristy83@gmail.com Dr.R.Arunkumar, Department of Computer Science and Engineering, Annamalai University, Chidambaram, Cuddalore, TamilNadu, India. Email: arunkumar_an@yahoo.com Though default approaches have been used for the past years, the 2007/2008 financial crisis lead to the effective FCP models with utmost priority. But, [2] suggested that no standard theories or models exists for corporate FCP. The absence of theoretical model to investigate financial crisis for exploratory actions for the identification of discriminant features and prediction models using trial and error [3][4]. The academicians and professionals wanted to enhance the performance of FCP models by the use of diverse quantitative models. For example, [5] developed the earliest logistic regression (LR) approach for default computation. Contrastingly, [1] provides a score to classify the observations as either good or bad customers; Ohlson‘s model computes the standard possibility of the significant. Assuming the relative ease of performing discriminant analysis and logistic regression, different works has been done to carry out identical tests. However, [6] disagreed that the famous Altman (1968) and Ohlson (1980) models are not precise and recommended the requirement of improvements in the modeling of default risks. Researchers discovered the artificial intelligence and ML approaches to measure credit risk using the recent technologies. As the investigation of financial crisis is identical to the pattern- recognition problems, methodologies can be employed for the classification of the creditworthiness, hence enhancing the conventional methods using earlier multivariate statistical methodologies like discriminant analysis and LR. Artificial neural networks (ANN) are also employed in various forms and the integration of ML algorithms in FCP is found to be interesting. Though numerous works has been investigated FCP by the use of recent techniques, [2] found that the results has not identifies the novel approach. More number of FCP models are developed using the conventional statistical models and early artificial intelligence models. The key facts of this investigation are to examine the generous change in forecast exactness utilizing ML strategies compared to statistical models. This paper performs a comparative analysis of deep learning (DL), multilayer Perceptron (MLP), radial basis function (RBF) network and logistic regression (LR). For evaluation, three benchmark dataset namely German dataset, Weislaw dataset and Polish dataset. From experimentation, it is reported that the DL based classifier outperforms the other algorithms in terms of various performance measures. II. RELATED WORKS FCP using the past history of the financial data is an interesting topic. Several works has been done on the domain of FCP [31].Discriminant analysis and Logit analysis are the widely used statistical models for FCP [32]. Altman Z-score [33] is most highly employed in this discriminant analysis.